Mine Water Environ (2012) 31:339–343
DOI 10.1007/s10230-012-0201-9
TECHNICAL COMMUNICATION
Effective Parameter Predictions in Metals Transport
from the Zanjan Zinc Mine Tailings using PHREEQC
Ali Motalebi Damuchali • Gholamreza Asadollahfardi
Ahmad Khodadadi
Received: 4 May 2012 / Accepted: 6 August 2012 / Published online: 17 August 2012
Ó Springer-Verlag 2012
Abstract The effects of grain size distribution, initial
flow rate, and initial pH were simulated using PHREEQC
to determine how these factors affect the transport of Cd,
Ni, and Mn from the tailings of the Zanjan zinc (Iran)
leaching plant. Not surprisingly, lower pH, higher flows,
and smaller effective particle size increased metal trans-
port. The simulation results generally agreed well with
results from a previously conducted column study.
Keywords Geochemical modeling  Sensitivity analysis 
Metal transport  PHREEQC  Mine tailing  Zanjan
Introduction
The Zanjan zinc leaching plant, located 12 km from Zan-
jan, Iran, processes ore from the affiliated Anguran mine
into zinc sheet metal. The tailings or ‘‘cake’’ from this
operation are enriched in nickel (Ni) and cadmium (Cd).
The objective of this study was to determine how pH, flow
rate, and grain size distribution affected Ni, Cd, and
manganese (Mn) transport from these tailings, using the
PHREEQC geochemical model.
Seuntjens et al. (2002) studied how physical and chemical
properties affected field-scale Cd transport in a heteroge-
neous soil profile; using Monte Carlo simulations, they
A. Motalebi Damuchali  G. Asadollahfardi (&)
Kharazmi University, Tehran, Iran
e-mail: asadollahfardi@yahoo.com; fardi@tmu.ac.ir
A. Motalebi Damuchali
e-mail: Motalebi@tmu.ac.ir
A. Khodadadi
Tarbiat Modares University, Tehran, Iran
e-mail: akdarban@modares.ac.ir
•
demonstrated that variations in field-scale Cd flux were
dominated by variations in deposition rate and the parame-
ters of the Freundlich sorption isotherm. Concas et al. (2005)
concluded that a decrease in pH increased metal dissolution
and transport from mine tailings. Tipping et al. (2006) used
the CHUM-AM model to investigate the behavior of atmo-
spherically-deposited metals in Cumbria, UK and concluded
that the principal processes controlling cationic metals are
competitive partitioning to soil organic matter, chemical
interactions in solution, and chemical weathering.
Michel et al. (2007) applied mathematical and empirical
models to investigate Ni and Cd transport in silty and sandy
soils and concluded that a decrease in pH increases metals
transport in acidic soils. Van der Grift and Griffoen (2007)
investigated Cd and Zn leaching from polluted soil using
unsaturated and saturated zone flow and reactive transport
models and concluded that metal contamination depended
on surface water load, geohydrologic and geologic struc-
ture, soil type, and dominant land use. Hanna et al. (2009)
used PHREEQC to study zinc and lead transport in soils
and concluded that the model can predict metal transport in
acidic conditions, and that decreasing pH increases metal
transport. Gordon et al. (2009) modeled Ni and Cd trans-
port and found that decreasing pH increases Ni and Cd
transport in silt and sandy soils. Horvath et al. (2009)
showed that column leaching tests can be effectively used
to investigate metal transport from mine tailings.
Materials and Methods
Simulation
PHREEQC.2 software (Parkhurst and Appelo 1999) was
used to simulate different scenarios of discharge, grain size
123
340
distributions, and pH for about 30 days. PHREEQC is a
geochemical model that is used for water and soil envi-
ronments. One dimensional transport with geochemical
reactions can be simulated. The modeling efficiency (EF)
was calculated for measured and modeled data using Eq. 1
(Loague and Green 1990):
Pn P
 xn Þ  ni¼1 ðyi  xi Þ
i¼1 ðxi P
EF ¼ n ð1Þ
i¼1 ðxi  xn Þ
where x, y, x-, and N are measured value, modeled pre-
diction, the arithmetic mean of measured values, and the
number of measured and modeled values, respectively.
When modeled predictions perfectly match the measured
ones, EF is equal to 1. A negative value indicates that the
modeled prediction describes the data less well than the
mean of the observations.
Ni–Cd Cake Compounds
The deposited waste material has been studied using X-ray
fluorescence (XRF) and by measurement of its physical
properties (Hashem Zadeh 2003; Sedaghat 2007). Elemental
analysis indicated that the material contains high concen-
trations of metals: 40.4 % zinc, 11.2 % Cd, 3.3 % Ni, 1.4 %
copper, and 0.8 % lead. A 1 kg sample was sieved using 18,
35, 60, 80, 100, 120, 140, 170, 200, 230, and 270 mesh sieves.
Sedaghat (2007) also conducted 30 day leaching tests on
1,441 g of the Ni–Cd cake in each of three 6 cm internal
diameter columns; the height of tailings in each column was
40 cm (of the 50 cm column height). The effluent of each
column was analyzed after 1, 3, 7, 13, 20, and 30 days. The
results of that study were used in this study. Parameters such
as rain events and acid leaching were simulated using
PHREEQC. Equilibrium and exchange constants for chem-
ical species of the tailings column were taken from the
MINTEQ data base (Allison et al. 1990).
Results and Discussion
Grain Size Distribution
Pore volumes differ with grain size distributions, so various
scenarios of grain size distribution were simulated in
PHREEQC. The first grain size distribution studied was a
mixed grain size distribution, since this type of distribution
exists in nature. The ?18 and -18 to ?60 grain size dis-
tributions represent a significant percentage (15 and 39 %,
respectively) of the Ni–Cd cake in the sieved analysis
(Sedaghat 2007). To investigate the effects of grain size
distribution on metal transport from the tailings, the pH of
the initial leaching solution was considered constant and
neutral. The initial solution flow rate was 1 mL/min. The
123
Mine Water Environ (2012) 31:339–343
cation species that participate in the exchange and equi-
librium reactions were defined by the software. The pre-
viously determined porosity coefficient for the mixed grain
size distribution was ?0.5 (Sedaghat 2007). Porosity cor-
responding to two other grain size distributions were cal-
culated using soil mechanics Eqs. 2, 3, 4, and 5 (Das 2008).
K ¼ 1  1:5D210 ð2Þ
where K and D10 is the permeability (cm/s) and effective
size (mm), respectively.
K ¼ 1:4e2 K0:85 ð3Þ
where K, e, and K0.85 permeability (cm/s), void ratio
(dimensionless), and the corresponding value of a void
ratio of 0.85 were calculated from Eq. (4):
  3 0:7825
e
K0:85 ¼ 2:4622 D210 ð4Þ
1þe
The relation between porosity (n) and e are defined in
Eq. (5):
e
n¼ ð5Þ
1þe
According to Eq. (5), the porosity for ?18 and -18 to
?60 grain size was 0.55 and 0.52, respectively.
Figure 1 compares the Cd concentration in the leachate
from various grain size distributions of tailings in the
PHREEQC simulation and the existing experimental data.
The Cd concentration decreased sharply until the 7th day
and then decreased at a much slower rate. The maximum
leaching concentration was in the mixed grain size distri-
bution and the minimum was associated with the ?18
fraction mesh. As indicated in Fig. 2, the Ni concentration
in the leachate behaved very similarly.
The model and experimental data for the Mn leachate
concentrations agree well for the -18 to ?60 and the ?18
grain size distribution, but for the mixed grain size distri-
bution, the model only agrees well with the experimental
data after the 7th day. In general, as the grain size distri-
bution was increased, the Mn concentration in in the
leachate decreased (Fig. 3).
Flow Rate
To investigate metals transport at various pH and flow rate
conditions using PHREEQC, we used one grain size dis-
tribution (mixed), since that distribution was the most
critical in metal transport and exists in nature (Sedaghat
2007). Three flow rates (low, moderate, and high intensi-
ties) were selected to simulate natural processes. A flow
rate of 0.5 mL/min produces a laminar flow and was the
lowest rate that could be reliably produced in our
Mine Water Environ (2012) 31:339–343 341

45 8
7.5
40 mix.mod 7

Mn Concentration (mg/L)
6.5
-18 to +60.mod +18.mod
6
Cd Concentration (mg/L)

35
+18.mod 5.5 -18 to +60.mod
5
30 mix.exp mix.mod
4.5
-18 to +60.exp 4 +18.exp
25 3.5
+18.exp -18 to +60.exp
3
20 2.5 mix.exp
2
15 1.5
1
10 0.5
0
0 5 10 15 20 25 30 35
5
Time (day)
0
0 5 10 15 20 25 30 35 Fig. 3 Simulation and measured manganese leaching concentrations
Time(day) with various grain size distributions

Fig. 1 Simulation and measured cadmium leaching concentrations

with various grain size distributions. Mod. and exp. represents model 45
and experiment respectively
40 0.5.mod

Cd Concentration (mg/L)
35 1.mod
2.mod
1.6 30
0.5.exp
25 1.exp
1.4 +18.mod
20 2.exp
-18 to +60.mod
15
Ni Concentration (mg/L)

1.2
mix.mod
10
1 +18.exp
5
-18 to +60.exp
0
0.8
mix.exp 0 5 10 15 20 25 30 35
Time (day)
0.6
Fig. 4 Modeled and measured leaching concentrations of cadmium at
0.4 various flow rates

0.2

1.6
0 0.5.mod
0 5 10 15 20 25 30 35 1.4
Ni Concentration (mg/L)

1.mod
Time (day)
1.2
2.mod
Fig. 2 Simulation and measured nickel leaching concentrations with 1 0.5.exp
various grain size distributions. Mod. and exp. represents model and 1.exp
0.8
experiment, respectively 2.exp
0.6

0.4
laboratory, while 2 mL/min was set as the highest condi-
tion since flooding occurred in the column at higher flow 0.2

rates (Sedaghat 2007). The maximum daily rainfall in 0

0 5 10 15 20 25 30 35
Zanjan is 55 mm/day (Islamic Republic of Iran Meteoro-
Time (day)
logical Organization 1983). The number of time steps was
defined for each flow rate in PHREEQC. The leachate Fig. 5 Modeled and measured leaching concentrations of nickel at
concentrations of Cd, Ni, and Mn for each flow rate are various flow rates
presented in Figs. 4, 5, and 6. For all three metals, leachate
concentrations from the tailings column decreased with steadily. The model results are compared to the results of
increasing initial flow rates and the amounts declined with the column leaching tests (Sedaghat 2007). The simulation
time during the 30 day period. Leachate concentrations efficiencies for the various grain size distributions and flow
plunged dramatically until the 7th day and then decreased rate scenarios are presented in Tables 1 and 2.

123
342 Mine Water Environ (2012) 31:339–343

Mn Concentration (mg/L) 12 Comparing Our Results with Other Research

10 0.5.mod Michel et al. (2007) investigated Ni and Cd removal in acidic

1.mod sandy and silty soils using empirical and mathematical
8
2.mod modeling. The simulation efficiency of their study was
0.5.exp between 0.35 and 0.85 % in different scenarios. They did not
6
1.exp compare results of modes with experiments and the effects of
2.exp chemicals or other parameters were not considered.
4
Hanna et al. (2009) studied the effects of increasing pore
2
water flow on zinc and lead transport in soil using
PHREEQC and observed weak compatibility between
0 predicted and measured results. They concluded that this
0 5 10 15 20 25 30 35 was because the software failed to consider the strong
Time (day)
metal-binding capacity of the soil.
Fig. 6 Modeled and measured leaching concentrations of manganese In this study, there were a good agreement between
at various flow rates model prediction and experimental data in most scenarios.
However, the proportion of leaching volume to tailings

Table 1 Simulation efficiency with various grain size distributions

60
Grain size distribution ?18 -18 to ?60 Mix
Cd Concentration (mg/L) 5.mod
Cadmium 0.82 0.93 0.61 50
6.mod
Nickel 0.36 0.45 0.3
40 7.mod
Manganese 0.49 0.84 0.44
5.exp
30 6.exp
7.exp
Table 2 Simulation efficiency at various flow rates 20

Flow rate 0.5 1 2

10
Cadmium 0.22 0.61 0.21
0
Nickel 0.62 0.3 0.2 0 5 10 15 20 25 30 35
Manganese 0.74 0.7 0.44 Time (day)

Fig. 7 Modeled and measured leaching concentrations of cadmium

over the pH range of 5–7
Initial pH

We used the same initial pH conditions (5, 6, and 7) in the 1.6

model as were used in the leaching study, which were 1.4 5.mod
intended to simulate the effects of acidic rainfall. The cation
Ni Concentration (mg/L)

species that would participate in the exchange and equilib- 1.2 6.mod

rium reactions were defined by the software. The reactive 7.mod

1
transport parameters were determined using a 1 mL/min 5.exp
flow rate, which represents moderate rainfall intensity in 0.8 6.exp
Zanjan (Sedaghat 2007), and a mixed grain size distribution.
0.6 7.exp
The number of time steps and the pore water flow velocity
were determined based on the observed tailings pore volume 0.4
and flow rate. The leaching concentrations of Cd, Ni and Mn
0.2
were predicted using PHREEQC for the three different pH
conditions after 1, 3, 7, 13, 20, and 30 days. 0
The leachate concentrations plunged dramatically with 0 5 10 15 20 25 30 35
Time (day)
time. Decreasing pH increased the solubility of all three
metals, as illustrated in Figs. 7, 8, and 9. The simulation Fig. 8 Modeled and measured leaching concentrations of nickel over
efficiencies for the pH scenarios are presented in Table 3. the pH range of 5–7

123
Mine Water Environ (2012) 31:339–343 343

10 concentrations of Cd, Ni, and Mn were associated with a

9 mixed grain size distribution, and the minimum was for a
Mn Concentration (mg/L)

5.exp
8 ?18 mesh grain size distribution. Generally, model pre-
6.exp
7 dictions for the effects of pH and grain size distribution
7.exp
6 variations were more compatible than the effects of flow
5.mod
5 rate variations. Of the three metals considered, the modeled
6.mod
4
7.mod
Ni concentrations were the least compatible in the different
3 scenarios. Incompatibility in some scenarios may be due to
2 an assumption by PHREEQC of linear cation retardation.
1
0
0 5 10 15 20 25 30 35
References
Time (day)

Fig. 9 Modeled and measured leaching concentrations of manganese
over the pH range of 5–7
Table 3 Simulation efficiency at various pH values
pH 5 6
Cadmium 0.91 0.93
Nickel 0.76 0.64
Manganese 0.98 0.97
column pore volume was high in all of the scenarios; this
may explain the observed differences between the results
of the predicted model and the column test. Poor agreement
in some scenarios could also be due to PHREEQC’s use of
a linear cation retardation factor. Retardation is an activity
of a cation species in an exchange reaction, which is
defined by Eq. 6 in PHREEQC:
1 þ CEC
R¼
C Reclam 17:415–424
where R is retardation, CEC is the cation exchange
capacity expressed in (Mol/kg water), and C is the species
concentration (Mol/kg water). In some scenarios, e.g. lower
pH, chemical reactions could compensate for the effects of
linear retardation and improve the compatibility between
modeled and measured data.
Conclusion
In general, metal transport predicted using PHREEQC for
various scenarios of grain size distribution, rainfall, and
acidity of simulated rainfall matched the results of the
laboratory column studies fairly well and illustrate that
geochemical modeling may be an alternative method to
investigate metals transport from mine tailings in various
environmental situations. In all cases, modeled effects of
increased grain size, lower flow rates, and higher initial pH
simulated the decreased concentration of metals leached
from the Zanjan tailings. The maximum simulated
Allison JD, Brown DS, Novo-Gradac KJ (1990) MINTEQA2/
PRODEFA2—A geochemical assessment model for environ-
mental system-version 3, user manual. Environmental Research
Laboratory, Office of Research and Development, US Environ-
mental Protection Agency, Athens
Concas A, Ardou C, Cristini A, Zuddas P, Cao G (2005) Mobility of
7 heavy metals from tailings to stream waters in a mining activity
contaminated site. Chemosphere 63:244–253
0.61 Das MB (2008) Advanced Soil Mechanics, 3rd edn. Taylor and
0.3 Francis, New York
0.44 Gordon GW, Lyons TW, Arnold GL, Roe J, Sageman BB, Anbar AD
(2009) When do black shales tell molybdenum isotope tales?
Geology 37:535–538
Hanna K, Lassabatere L, Bachet B (2009) Zinc and Lead transfer in a
contaminated road soil: experimental study and modelling.
J Hazard Mater 161:1499–1505
Hashem Zadeh M (2003) Study of the machinery methods for
quantity and quality analysis of mineral. MSc thesis, Tarbiat
Modares Univ, Tehran, Iran
Horvath E, Jordan G, Fugedia U, Bartha A, Kuti L, Heltai G, Kalmar
J, Wald I, Naperdean I, Jordan G, vanRompaey A, Somody A,
Fugedi U, Bats M, Farsang A (2009) Spatial modelling of
contamination in a catchment area impacted by mining: a case
ð6Þ study for the Recsk copper mines, Hungary. J Land Contam
Islamic Republic of Iran Meteorological Organization (1983) Climate
statistics. http://www.irimo.ir[. Accessed Jan 18 2012
Loague K, Green RE (1990) Criteria for evaluating pesticide leaching
models. J Contam Hydrol 7:51–73
Michel K, Roose M, Ludwig B (2007) Comparison of different
approaches for modelling heavy metal transport in acidic soils.
Geoderma 140:207–214
Parkhurst LB, Appelo CAJ (1999) User guide to PHREEQC (version 2)
A computer program for speciation, batch reaction, one dimen-
sional transport, and inverse geochemical calculation. USGS
Water-Resources Investigation Report 99–4259, Washington DC,
USA
Sedaghat B (2007) Investigation of effective parameters for hazardous
material transport from zinc leaching plants to environment.
MSc thesis, Tarbiat Modares Univ, Tehran, Iran
Seuntjens P, Mallants D, Simunek J, Patyn J, Jacques D (2002)
Sensitivity analysis of physical and chemical properties affecting
field—scale transport in a heterogeneous soil profile. J Hydrol
264:185–200
Tipping E, Lawlor AJ, Lofts S, Shotbolt L (2006) Simulating the long
term chemistry of an upland UK catchment: heavy metals.
Environ Pollut 141:139–150
van der Grift B, Griffoen J (2007) Modelling assessment of regional
ground water contamination due to the historic smelter emission
of heavy metals. J Contam Hydrol 96:48–68

123